AJNR Am J Neuroradiol 19:743–747, April 1998

Terson Syndrome: CT Evaluation in 12 Patients

Charles E. Swallow, Jay S. Tsuruda, Kathleen B. Digre, Matthew J. Glaser, H. Christian Davidson, and H. Ric Harnsberger

PURPOSE: Terson syndrome may be overlooked in the acute setting and often requires ophthalmologic intervention to prevent long-term visual loss. In this syndrome, vitreous or retinal hemorrhage results from an abrupt rise in intracranial pressure, leading to retinal venous and . Our objective was to determine whether imaging findings could be discovered that might facilitate an earlier diagnosis. METHODS: Our inpatient medical record data base for 1991–1996 listed 11 patients with Terson syndrome. The medical records of these 11 patients were reviewed retrospectively and compared with their noncontrast head CT scans and with scans of 10 control subjects. One additional case was discovered prospectively, for a total of 12 patients. Three radiologists unaware of the patients’ history evaluated CT scans of the orbits for evidence of intraocular hemorrhage. RESULTS: CT findings in eight patients were suggestive of retinal hemorrhage manifested by a retinal crescent or nodule that was slightly hyperdense relative to the vitreous humor. There was a high degree of concordance between the retrospective and independent reviews. CONCLUSION: Retinal nodularity and crescentic hyperdensities are evident on CT scans in the majority of patients with Terson syndrome. Although findings are subtle and not present in all cases, in the setting of they suggest retinal hemorrhage and warrant detailed fundoscopic evaluation.

Terson syndrome is a vitreous or retinal hemor- whether specific findings of this entity could be iden- rhage occurring consequent to subarachnoid hemor- tified on CT studies of the brain. rhage; it is strongly associated with visual deficits and may require early surgical management (1–4). Fun- doscopic examination is the usual means of diagnosis, but it may be difficult to perform or the may Methods be overlooked in the setting of acute subarachnoid hemorrhage. The presence of Terson syndrome has The inpatient medical records from the computer data base been reported to be predictive of poor clinical out- for the period 1990 through 1996 were searched for patients discharged with the diagnosis of Terson syndrome or with come in terms of recovery from the intracranial hem- diagnostic combinations of subarachnoid hemorrhage and vit- orrhage (4–6). We reasoned that if characteristic reous or retinal hemorrhage. The search generated 11 patients. imaging findings of Terson syndrome could be estab- The imaging studies and medical records of these patients were lished, earlier diagnosis could prove useful as a prog- reviewed retrospectively by two radiologists. A subsequent case nostic sign. We therefore reviewed CT scans in 12 (case 8, see Table) was diagnosed prospectively on the basis of CT findings. patients discharged from our facility with the diagno- The records of the 12 patients with Terson syndrome and sis of Terson syndrome in an effort to determine those of 10 control subjects with intracranial hemorrhage but without Terson syndrome as noted on fundoscopic examination were reviewed by three radiologists (a senior radiology resi- dent, a neuroradiology fellow, and a neuroradiology professor). Received June 13, 1997; accepted after revision October 1. The reviewers were unaware of which subjects had ocular Presented at the annual meeting of the American Society of disease, and were asked to judge the retinal surfaces as normal, Head and Neck Radiology/American Society of Neuroradiology, abnormal, or obscured. Toronto, Canada, May 1997. From the Departments of Radiology (C.E.S., J.S.T., M.J.G., The majority of imaging studies were performed with a 23-cm field of view, 120 kV, 170 mA, 2-second scan time, and H.C.D., H.R.H.) and (K.B.D.), University of Utah ϫ Health Science Center, Salt Lake City. 512 512 matrix, and photographed at a window width of 150 Address reprint requests to Jay S. Tsuruda, MD, Department of and level of 35 (GE 9800 or High Speed Advantage). In a single Radiology, Room A171, 50 N Medical Center Dr, Salt Lake City, patient, imaging parameters included a 22-cm field of view, 130 UT 84132. kV, 125 mA, 3-second scan time, and 512 ϫ 512 matrix, pho- tographed at a window width of 80 and level of 40 (Picker © American Society of Neuroradiology 1200).

743 744 SWALLOW AJNR: 19, April 1998

Terson syndrome: data in 12 cases

Days to Cause of Documented Eye Outcome Time to Case Age, y Retrospective CT Findings Abnormal Hemorrhage Examination Visual Acuity Diagnosis CT Findings

1 46 ACOM Vitreous and retinal heme, L crescentic density temporal Bilateral 2wk 12 OS Ͼ OD to macula 20/20 OS, 20/30 OD 2 29 Giant L MCA Central retinal and 3 ϫ 3-mm L macula nodule Vitrectomy OS, 20/20 3wk 2 aneurysm vitreous heme OS OU 3 35 Trauma Vitreous and retinal 8 ϫ 3-mm hyperdense crescent 20/15, 20/25 3 mo 1 hemorrhage OS temporal to L ON 4 42 ACOM Vitreous and retinal R temporal retinal thickening Unknown Unknown 1 aneurysm hemorrhage OD and nodularity near ON 5 60 ACOM OD ϾϾ OS vitreous and L hyperdense crescent medial Unknown 2 mo 20 aneurysm retinal hemorrhage retinal surface 6 20 Trauma Bilateral vitreous and Bilateral 5 ϫ 15-mm central Death Immediate 1 retinal hemorrhage retinal hyperdense crescents 7 16 Trauma Central and inferomedial R crescentic density central Unknown Immediate 1 retinal heme OD and inferior 8 55 ACOM Temporal retinal R retinal crescentic nodule Unknown 2 d 2 aneurysm hemorrhage OD temporal to ON 9 45 LVA aneurysm, Small dot blot foci of Normal retina Spontaneous 3wk ... cocaine retinal heme OS clearance improved to 20/80 OS 10 50 L PICA Vitreous heme OS, small Normal retina Spontaneous 3mo ... aneurysm macular heme clearance improved to 20/70 OS 11 57 Undetermined Small retinal heme inferior Normal retina Never had visual Unknown . . . to disk OD deficit 12 72 ACOM rupture Retinal heme OD Normal retina Unknown Unknown . . . 3

Note.—ACOM indicates anterior communicating artery; MCA, middle cerebral artery; LVA, left vertebral artery; PICA, posterior inferior cerebellar artery; OS, left eye; OD, right eye; ON, .

Results that time showed retinal hemorrhage. His earlier studies were not available for review. In one patient The results of the imaging and chart reviews are (case 9) imaged on an older-model scanner, evidence summarized in the Table. Intracranial hemorrhage resulted from aneurysmal rupture in eight cases and of retinal hemorrhage was not appreciated until seen from head trauma in three cases. In one case, the on scans obtained 20 days after the initial intracranial cause of the hemorrhage could not be determined hemorrhage. despite multiple angiograms. Although the CT findings ranged from subtle asym- The degree of vitreous and retinal hemorrhage metric retinal thickening to dense crescents 3 to 5 mm documented in the ocular examination varied sub- thick, typical findings consisted of a crescent or nod- stantially among patients, ranging from extensive ule on the retinal surface that was slightly hyperdense hemorrhage in some patients to only minimal findings relative to the vitreous (Figs 1–3). The clinically doc- in others. Terson syndrome was diagnosed in 15 umented hemorrhage in the vitreous was often large, globes examined fundoscopically. The time to oph- to the point of obscuring the retina, but no qualitative thalmologic diagnosis varied from immediate diagno- difference could be discerned between the involved sis in three cases to several weeks to months in half and uninvolved vitreous on CT scans. Hounsfield unit the cases, often related to discovery of a late visual measurements were obtained for the vitreous humor defect hindering physical therapy or rehabilitation. in the first five patients with Terson syndrome and in Retrospective review of the CT scans revealed find- five control patients, but no significant difference in ings suggestive of retinal hemorrhage in eight of the the average vitreous density was found between Ter- patients. This included nine of the 15 globes with son patients and control subjects or in the density of hemorrhage on physical examination. In six of these the involved and uninvolved vitreous in individual patients, findings suggestive of retinal hemorrhage patients despite the presence of were discovered on CT scans obtained within 2 days documented on fundoscopic examination (Fig 2). Be- after intracranial hemorrhage. The hemorrhage be- cause no significant differences in vitreous density came slightly more dense and conspicuous over the were found, no Hounsfield unit measurements were subsequent week in three of those cases. In one pa- obtained in the subsequent cases. tient (case 1) transferred to our facility 12 days after The retrospective review of the images by the three his intracranial hemorrhage, an admission CT scan at radiologists was in agreement with findings in the AJNR: 19, April 1998 TERSON SYNDROME 745

FIG 1. Case 3: 35-year-old woman with severe closed head FIG 3. Case 2: 29-year-old woman 12 days postpartum suf- after being struck by a car. Two months later a left visual fered rupture of a giant fusiform left middle cerebral artery an- field deficit was noted during admission for shunt revision. Fun- eurysm. A visual field deficit was noted 3 weeks later, at which doscopy revealed retinal hemorrhage in the temporal pole and time fundoscopy revealed central retinal and vitreous hemor- old vitreous blood in the left eye. A CT scan obtained on the day rhage in the left eye. A CT scan obtained on the day after the of the trauma revealed a hyperdense crescent (arrow)inthe intracranial hemorrhage revealed a hyperdense nodule (arrow) temporal retinal surface adjacent to the optic nerve correspond- on the central left retinal surface corresponding to the fundo- ing to the site of hemorrhage on fundoscopy. scopic abnormality.

false-positive diagnoses, the majority were found to be due to mistaking the tangentially imaged superior or inferior aspects of the for a retinal abnor- mality. In two cases, a streak artifact and in one case a rectus muscle insertion were mistaken for a retinal hemorrhage. Knowledge of the clinical outcome was available in six patients. In two cases, a pars plana vitrectomy was performed, resulting in improved visual acuity. In case 1, vision improved from “count fingers” to 20/20 in the left eye and to 20/30 in the right eye. Case 2 underwent a left vitrectomy resulting in 20/20 visual acuity; however, the preoperative acuity was not FIG 2. Case 4: 42-year-old man after aneurysmal rupture and noted in the inpatient records. In case 11 the small clipping in the anterior communicating artery. Physical examina- hemorrhages never resulted in a significant visual tion revealed vitreous and retinal hemorrhage in the right eye. A CT scan obtained 1 week after intracranial hemorrhage revealed deficit, and in cases 9 and 10 spontaneous clearing of a hyperdense crescent (arrow) on the temporal retinal surface the vitreous hemorrhage resulted in improved visual adjacent to the optic nerve corresponding to hemorrhage on acuity from 20/400 to 20/80 and from 20/400 to 20/70, fundoscopy. Note the apparent increased density of the normal respectively. Patient 6 died shortly after his injury. left vitreous as compared with the abnormal right side. control subjects in the majority of cases. The most Discussion experienced reviewer achieved the highest rate of Terson syndrome is defined as vitreous or retinal concordance (11 of 12). There was agreement with hemorrhage associated with subarachnoid hemor- the retrospective review in 10 of 12 cases reviewed by rhage and is thought most likely to result from a the fellow, and in nine of 12 cases reviewed by the sudden large increase in intracranial pressure (1). It is resident. In the single case in which the three inde- seen in 10% to 20% of patients with spontaneous or pendent reviewers all disagreed with the retrospective traumatic subarachnoid hemorrhage and generally analysis (case 5), the findings were subtle, partially begins as bleeding between the internal limiting mem- obscured by streak artifact, and had been imaged on brane and the retina (3, 7). The hemorrhage may be the oldest model CT scanner. In case 10, the fellow localized to this site or may extend into the subhyloi- predicted an abnormality that corresponded with ret- dal or vitreous spaces. The pathogenesis of the ocular inal hemorrhage on physical examination, although findings is thought to be related to disturbed circula- no lesion was detected on either of the retrospective tion in the retinal vessels consequent to the increased reviews. None of the control studies were called ab- intracranial pressure, which leads to retinal venous normal by the professor, although in one case of hypertension and, eventually, to hemorrhage (8–10). unilateral Terson syndrome (case 8), bilateral lesions The clinical course of this process is variable. While in were described. Three control cases were called ab- some patients, the hemorrhage clears spontaneously, normal and three were called obscured by the fellow. many will incur vision loss, chronic hematoma, or One control study was called abnormal and two were epiretinal membranes requiring ophthalmologic care, called obscured by the resident. In reviewing the including vitrectomy (3, 7, 8). 746 SWALLOW AJNR: 19, April 1998

Some studies have suggested that patients with in- the blood products are cleared over a period of traocular hemorrhage in the setting of acute sub- months by macrophages, that large hemorrhages are arachnoid hemorrhage have a higher mortality rate associated with poor clinical outcome, and that some (50% to 90%) than do patients without Terson syn- of the damage to the retina may be related to toxic drome, and that the presence of intraocular blood effects of iron in large hemorrhages (7, 8, 12). Vitre- may be a manifestation of rebleeding from the aneu- ous hemorrhage, the principle fundoscopic finding, rysm (4, 6). In another series, no difference in mor- could not be identified on CT scans. During the eval- tality was found between patients with and patients uation of the first five patients in this series, an at- without intraocular hemorrhage (5). Only one patient tempt was made to compare the vitreous Hounsfield in our group of 12 died. This comparatively low mor- unit measurements in the patients with Terson syn- tality is most likely a result of case selection bias, since drome against those in the control subjects, but no the retrospective search parameters we used would significant density differences between patients and favor patients who survived to long-term follow-up control subjects or between normal and abnormal and ophthalmologic evaluation. globes in the same patient could be found. This was Clinical examination and pathologic studies of the thought to be due to the baseline increased attenua- hemorrhages in Terson syndrome have revealed a tion of the vitreous humor relative to cerebrospinal dome-shaped posterior pole membrane arising in the fluid, which, combined with streak artifacts commonly subhyaloid space, typically in the temporal vascular found over the globes, limits CT contrast resolution arcade (6, 11). The blood may remain confined in this between the radiologically dense vitreous and the space and appear as a mound on the retina, or may similarly dense intermixed blood products. partially or completely decompress through the hya- No CT evidence of ocular abnormalities could be loid membrane and seep into the vitreous (Fig 4). We found in four of the clinically positive cases in this postulate that the crescents seen on our imaging stud- series. In some patients, these false-negative studies ies represent the subhyaloid blood. may have been due to the small size of the irregularity To our knowledge, the imaging features of Terson on the retinal surface. In others, the lower image syndrome and the evolution of retinal hemorrhage quality of the earlier model CT scanners may have have not been previously described. As would be played a role. The subtle findings we have described expected from the pathologic studies described may be obscured or difficult to appreciate owing to above, the presence of retinal hemorrhage may be streak artifacts from the adjacent orbits or to motion identified on CT scans. In our series, positive findings artifacts in acutely ill patients. A potential cause of a on CT scans included thickening, nodularity, or cres- false-positive CT finding is volume averaging at the cent-shaped increased density of the retinal surface, rectus muscle insertions or tangential sectioning of often adjacent to the optic nerve and along the tem- the superior and inferior aspects of the globe, which poral surface of the retina. These findings were iden- may mimic a crescentic increased density in the ret- tified within the first few days after intracranial hem- ina. The true findings might be most reliably appre- orrhage in the majority of patients with CT evidence ciated on the surface of the retina adjacent to the of Terson syndrome. In three cases, the retinal hem- optic nerve, where volume averaging is less likely to orrhages became slightly more dense over time, pos- occur. Reduction of the section profile from 5 mm to sibly related to consolidation and retraction of the 3 mm thickness might decrease the volume-averaging thrombus. Clinical studies involving patients with artifacts, but such a section profile is not routinely Terson syndrome as well as subretinal hemorrhage used when screening for subarachnoid hemorrhage or related to senile suggest that in subsequent follow-up studies when ocular disease is not suspected. Although the differential diagnosis of a noncalci- fied retinal nodule would include such entities as melanoma, metastasis, and hemangioma, in the set- ting of acute subarachnoid hemorrhage, where there is a high prevalence of retinal hemorrhage, Terson syndrome is the most likely possibility. In either case, the detection of a retinal nodule should lead to fur- ther detailed evaluation.

Conclusion Terson syndrome is a relatively common complica- tion of subarachnoid hemorrhage that may require early surgical attention to prevent long-term vision loss and hasten recovery (2, 3, 8). Although the diag- nosis is typically made fundoscopically, this examina-

FIG 4. Illustration of hemorrhage (large arrow) arising between tion may be limited in an acutely ill patient or findings the retina (small arrow) and the internal limiting membrane (ar- may evolve after the initial evaluation. The diagnosis rowhead) with decompression into the vitreous. may not be made until weeks or months into rehabil- AJNR: 19, April 1998 TERSON SYNDROME 747 itation, when a visual field deficit is discovered. We Terson syndrome in spontaneous subarachnoid hemorrhage, a have described findings of retinal nodularity and cres- prospective study in 60 consecutive patients. J Neurosurg 1996;85: 392–394 centic increased density in two thirds of the patients 5. Garfinkle AM, Danys IR, Nicole DA. Terson syndrome: a revers- in this small series, with a high concordance rate ible cause of blindness following subarachnoid hemorrhage. J Neu- between the retrospective and independent reviews. rosurg 1992;76:766–771 Although the findings are often subtle and not 6. Shaw HE, Landers MB. Vitreous hemorrhage after intracranial hemorrhage. Am J Ophthalmol 1975;80:207–213 present in all patients, discovering them in the setting 7. Williams DF, Mieler WF, Williams GA. Posterior segment mani- of subarachnoid hemorrhage suggests the presence of festations of ocular trauma. Retina 1990;10(Suppl 1):S35–S44 retinal hemorrhage and should prompt a detailed 8. Velikay M, Datlinger P, Stolba U, Wedrich A, Binder S, Hausmann fundoscopic examination. N. with severe proliferative vitreoretinopathy in Terson syndrome. Ophthalmology 1994;101:35–37 9. Weingeist TA, Goldman EJ, Fulk JC, Packer AJ, Ossoinig KC. Terson syndrome: clinicopathologic correlation. Ophthalmology References 1986;93:1435–1442 1. Kincaid MC, Yanoff M, Fine BS. Vitreous. In: Duane’s Foundations 10. Smith DC, Kearns TP, Sayre GP. Preretinal and optic nerve sheath of Clinical Ophthalmology 95, vol 3. Lippincott-Raven; 1995:4–5 hemorrhage: pathologic and experimental aspects in subarachnoid 2. Schultz PN, Sobol WM, Weingeist TA. Long-term visual outcome hemorrhage. Trans Am Acad Ophthalmol Otolaryngol 1957;61:201– in Terson syndrome. Ophthalmology 1991;98:1814–1819 210 3. Roux FX, Panthier JN, Tanghe YM, et al. Terson syndrome and 11. Vanderlinden RG, Chisholm LD. Vitreous hemorrhage and sudden intraocular complications in meningeal hemorrhages (26 cases) increased intracranial pressure. J Neurosurg 1974;41:167–176 Neurochirurgie 1991;37:106–110 12. Lim JI. Subretinal hemorrhage. Int Ophthalmol Clin 1995;35:95– 4. Pfausler B, Belcl R, Metzler R, Mohsenipour I, Schmutzhard E. 104